CFP-SAEF: A Spatial-Attention Enhanced Feature Fusion Network for Carbon Fiber Prepreg Defect Detection

CFP-SAEF: A Spatial-Attention Enhanced Feature Fusion Network for Carbon Fiber Prepreg Defect Detection | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article CFP-SAEF: A Spatial-Attention Enhanced Feature Fusion Network for Carbon Fiber Prepreg Defect Detection Jiazhong Xu, Yuteng Yue, Hongyi Guo, Kewei Sun This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7499409/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Carbon fiber reinforced polymer (CFRP) composites are widely used in aerospace, automotive manufacturing, and new energy equipment due to their high specific strength, lightweight, and corrosion resistance. However, during the manufacturing process of carbon fiber prepreg, small-sized, morphologically diverse, and multi-scale surface defects—such as cracks, voids, wrinkles, and foreign matter—frequently occur. If not detected in time, these defects can severely compromise the structural strength and service life of CFRP components. To address this issue, this paper proposes a carbon fiber prepreg defect detection network based on spatial-attention enhanced feature fusion (CFP-SAEF). The proposed method employs EfficientNet and UniRepLKNet as complementary backbones to form a dual-encoder structure, integrates their multi-scale features through a Multi-stage Adaptive Feature Fusion Module (MAFFM), and incorporates a Spatial Attention Enhanced Feature Pyramid Network (SAE-FPN) to selectively strengthen responses to small defect regions. Experimental results demonstrate that, on the carbon fiber prepreg defect dataset, CFP-SAEF achieves an mAP of 93.82%, outperforming YOLOv8m by 3.61 percentage points, with AP gains of 4.1% and 5.6% for the slit and foreign matter categories, respectively. Furthermore, on the PASCAL VOC2007 and KolektorSDD datasets, CFP-SAEF also attains higher detection accuracy compared with existing methods, validating the effectiveness and generalization capability of the proposed approach. carbon fiber prepreg defect detection multi-scale feature fusion spatial attention Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 03 Sep, 2025 Editor assigned by journal 01 Sep, 2025 Submission checks completed at journal 01 Sep, 2025 First submitted to journal 31 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7499409","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":509454914,"identity":"515c7442-7b35-4762-b274-03274b0e1704","order_by":0,"name":"Jiazhong Xu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3klEQVRIie2RMQrCQBBFB4TYRLaTDYpeYSWFCB5mgq0HSGkhVjlAxCtYpLScZQubSNqAFhHBSkGxsUhhotZm7QT3MTDNf3yGATAYfhMkgGKgRnTxv1MsT4axflOp2K5qTDWyIsU93fNdp8+WF2pMoMua9FlxQkQZ2Ed3EJ4jclbQmy/ws8I4ItlceVG6iagXA4pthWIVisxFqcQZeTMNpWxRNhZKEgBJHcUJMlRtUq5ILSEnMa++RazHo+spVx2RqMMt94dd1qpQine8E/y5eVW8pE6vzUgnbTAYDP/IA/J0VCkLKnJjAAAAAElFTkSuQmCC","orcid":"","institution":"School of Mechanical and Power Engineering, Harbin University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Jiazhong","middleName":"","lastName":"Xu","suffix":""},{"id":509454916,"identity":"14ec235b-5e11-4d67-ba15-f5268a947d8c","order_by":1,"name":"Yuteng Yue","email":"","orcid":"","institution":"Harbin University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Yuteng","middleName":"","lastName":"Yue","suffix":""},{"id":509454917,"identity":"dc5687ea-005f-412f-b041-1c61de73cffb","order_by":2,"name":"Hongyi Guo","email":"","orcid":"","institution":"Harbin University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Hongyi","middleName":"","lastName":"Guo","suffix":""},{"id":509454919,"identity":"e2044471-c93c-4cfc-a57d-b967d44f8b03","order_by":3,"name":"Kewei Sun","email":"","orcid":"","institution":"Weihai Research Institute, Harbin University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Kewei","middleName":"","lastName":"Sun","suffix":""}],"badges":[],"createdAt":"2025-08-31 08:53:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7499409/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7499409/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":90869878,"identity":"a07a760f-ac89-4ed8-82d0-576dc1834ce9","added_by":"auto","created_at":"2025-09-09 08:02:22","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1331479,"visible":true,"origin":"","legend":"","description":"","filename":"CFPSAEFASpatialAttentionEnhancedFeatureFusionNetworkforCarbonFiberPrepregDefectDetection.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7499409/v1_covered_22266fd1-dcc3-4fe1-b765-a65148479d8a.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"CFP-SAEF: A Spatial-Attention Enhanced Feature Fusion Network for Carbon Fiber Prepreg Defect Detection","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"signal-image-and-video-processing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sivp","sideBox":"Learn more about [Signal, Image and Video Processing](http://link.springer.com/journal/11760)","snPcode":"11760","submissionUrl":"https://submission.nature.com/new-submission/11760/3","title":"Signal, Image and Video Processing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"carbon fiber prepreg, defect detection,multi-scale feature fusion, spatial attention","lastPublishedDoi":"10.21203/rs.3.rs-7499409/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7499409/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Carbon fiber reinforced polymer (CFRP) composites are widely used in aerospace, automotive manufacturing, and new energy equipment due to their high specific strength, lightweight, and corrosion resistance. However, during the manufacturing process of carbon fiber prepreg, small-sized, morphologically diverse, and multi-scale surface defects—such as cracks, voids, wrinkles, and foreign matter—frequently occur. If not detected in time, these defects can severely compromise the structural strength and service life of CFRP components. To address this issue, this paper proposes a carbon fiber prepreg defect detection network based on spatial-attention enhanced feature fusion (CFP-SAEF). The proposed method employs EfficientNet and UniRepLKNet as complementary backbones to form a dual-encoder structure, integrates their multi-scale features through a Multi-stage Adaptive Feature Fusion Module (MAFFM), and incorporates a Spatial Attention Enhanced Feature Pyramid Network (SAE-FPN) to selectively strengthen responses to small defect regions. Experimental results demonstrate that, on the carbon fiber prepreg defect dataset, CFP-SAEF achieves an mAP of 93.82%, outperforming YOLOv8m by 3.61 percentage points, with AP gains of 4.1% and 5.6% for the slit and foreign matter categories, respectively. Furthermore, on the PASCAL VOC2007 and KolektorSDD datasets, CFP-SAEF also attains higher detection accuracy compared with existing methods, validating the effectiveness and generalization capability of the proposed approach.","manuscriptTitle":"CFP-SAEF: A Spatial-Attention Enhanced Feature Fusion Network for Carbon Fiber Prepreg Defect Detection","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-09 07:54:16","doi":"10.21203/rs.3.rs-7499409/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2025-09-03T06:04:42+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-01T17:16:33+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-01T17:16:06+00:00","index":"","fulltext":""},{"type":"submitted","content":"Signal, Image and Video Processing","date":"2025-08-31T08:51:48+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"signal-image-and-video-processing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sivp","sideBox":"Learn more about [Signal, Image and Video Processing](http://link.springer.com/journal/11760)","snPcode":"11760","submissionUrl":"https://submission.nature.com/new-submission/11760/3","title":"Signal, Image and Video Processing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"e1e5924a-b1a4-4768-95f6-8da01ca2681b","owner":[],"postedDate":"September 9th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-09-09T07:54:16+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-09 07:54:16","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7499409","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7499409","identity":"rs-7499409","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}